MLOps before DLS?


the “logical” continuation after the MLS would be the DLS. However, after nearly having finished the MLS, I am rather curios how all this works “in production”, so I am considering to do the MLOps first and then the DLS.

Does this also make sense or are there strong reasons to do the DLS first and only then the MLOps?

Thanks for any hint

Hey @Matthias_Kleine,
I believe you can audit the courses of both the specialization to see which one, in your opinion, should you do first. But let me provide my opinion as well. However, I would like you to note that I have done the MLS and DLS, but not the MLOps Specialization yet.

I believe that the MLOps Specialization, will be more focused towards teaching how to deploy models rather than teaching how are models built and how they work under the hood. So, if in the MLOps Specialization, say they teach how to deploy a deep-learning based model, they won’t be teaching how the model works, and the different aspects of the model, which could frame how we form the pipeline for that model.

In that case, you would have to search through the Web to learn about such models, the chances for which will decrease by manifolds, if you would have already done the DLS.

Allow me to quote the description from the Specialization:

Understanding machine learning and deep learning concepts is essential, but if you’re looking to build an effective AI career, you need production engineering capabilities as well.

So, I believe they expect learners to at least know the basics of deep learning. Now, although Prof Andrew have covered some very basic elements of Deep Learning in the MLS, but I don’t think those would be enough. Plus if you are not on a timeline, then there is no harm in doing DLS first, if learning Deep Learning is among one of your goals.

So, I believe completing DLS would be better than before pursuing MLOps.